MySQL "INSERT INTO SELECT NOT EXISTS" Efficiency - mysql

Question

The following bit of code is causing me no end of headaches, it works; but it is taking upwards of 30min with only 16k rows; the final database will be looking to include 1-5M rows.
So I'm looking to make things more efficient
INSERT INTO r (aKey, bKey, c, d, e, fKey, g, h, i, j, k, l, mKey, nKey, o, pKey, qKey, r, sKey)
SELECT aKey, bKey, c, d, e, fKey, g, h, i, j, k, l, mKey, nKey, o, pKey, qKey, r, sKey
FROM t2
WHERE NOT EXISTS
(
SELECT i,g
FROM r
WHERE t2.i = r.i AND t2.g = r.g
);
The field of i & g are the uniques (not specified in the DB currently - should they be? but a requirement)
Anything with Key on is a foreign key referencing another table
After the above statement I then loop through each none unique and update if it is different.
UPDATE r
INNER JOIN t2 ON t2.i = r.i
SET r.d=t2.d;
I have looked into :
INSERT ... ON DUPLICATE KEY UPDATE
but I was never able to get it to work; as all the examples I found were not dealing with a insert select
Any suggestions to improve the SQL will be most helpful.

Answer 1

Conclusion was: Due to the size of the request best to go back to basics and "set unique database fields to UNIQUE" and then do a INSERT IGNORE
Code below: (including python variable inserts)
INSERT IGNORE INTO """ + tblName + """ (""" + dupeString + """)
SELECT """ + dupeString + """
FROM t2
ON DUPLICATE KEY UPDATE """ + updateString + """;
dropped from 30-40m query to... less than 3 seconds.